From: Degerstrom, Michael J. ([email protected])
Date: Tue Jun 05 2001 - 07:28:56 PDT
We are designing a board with many very high I/O
BGA parts. In some cases, the I/O strength/style
are selectable but there are many cases where
output impedances are not selectable and have
very low impedances of say 10-15 ohms. In the
past, with lower pin counts and less dense I/O,
very good signal integrity was obtained by adding
30-40 ohm series (source) termination resistors close to
the driving pins. But using resistors is not very
practical with this design due to the high I/O
count with multiple BGA packages separated by
only about .200".
Without resistors, the ringing (overshoot and then
ringback)is substantial. In some cases I wouldn't
be so concerned but many of the inputs are "LVTTL"
in which case the vendor MAY bias the threshold
closer to ground. Of course, in many cases, these
inputs are driven by "LVTTL" outputs which pull down
much harder than they pull-up and so the LOW state
ringing is higher than the HIGH state ringing.
Having said all this I'm hoping that some of you
can respond to me with some guidelines, techniques,
etc., that you use to obtain successful "resistorless"
designs. Here are some things that I have been
1) Don't worry about ringing unless the signals
haven't stablized before the input device is
2) Rely on the clamps in the inputs to reduce
ringing and protect them from overvoltage.
3) Trust everything to the IBIS models - model
output to input(s) - if there are no glitches
then everything will work.
4) Lower the board impedance to 35-40 ohms to
reduce ringing - but what about increased power and
5) Terminate the top 20% "worst offenders".
6) Pick overshoot and ringback maximum values
and try to somehow design the nets to meet these
values. If don't think the parts can be placed
any closer however. What max. values of overshoot and
ringback do I chose?
Thanks in advance,
Mike Degerstrom Email: [email protected]
Mayo Clinic; 200 1st Street SW ; Rochester, MN 55905
Phone: (507) 538-5462 FAX: (507) 284-9171
**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected] In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
This archive was generated by hypermail 2b29 : Thu Jun 21 2001 - 10:12:15 PDT